Download "Torque-Balanced High-Density Steady States of Single-Component Plasmas" Physical Review Letters 94 (2005), 0305001. J. R. Danielson and C. M. Surko (PDF)

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts

Mathematics of radio engineering wikipedia, lookup

Chirp spectrum wikipedia, lookup

Utility frequency wikipedia, lookup

Spectral density wikipedia, lookup

Dynamic range compression wikipedia, lookup

Heterodyne wikipedia, lookup

Plasma display wikipedia, lookup

Transcript
PRL 94, 035001 (2005)
PHYSICAL REVIEW LETTERS
week ending
28 JANUARY 2005
Torque-Balanced High-Density Steady States of Single-Component Plasmas
J. R. Danielson and C. M. Surko
Department of Physics, University of California at San Diego, La Jolla, California 92093, USA
(Received 10 September 2004; published 24 January 2005)
Electron plasmas in a Penning-Malmberg trap are compressed radially using a rotating electric field (the
‘‘rotating-wall technique’’). For large electric fields, plasmas can be compressed over a broad range of
frequencies. This permits access to a novel high-density regime in which outward transport is insensitive
to plasma density. The limiting density occurs when the plasma rotation frequency equals the rotating-wall
frequency. Characteristics of the resulting torque-balanced steady states are described, and implications
for high-density electron and positron plasma confinement are discussed.
DOI: 10.1103/PhysRevLett.94.035001
PACS numbers: 52.27.Jt, 52.25.Fi, 52.55.Dy
Trapped single-component plasmas are important in
many contexts, such as the confinement and tailoring of
antiproton and positron plasmas for numerous applications, including atomic physics and tests of fundamental
symmetries and the characterization of materials [1].
Consequently, it is important to understand the range of
accessible plasma parameters, including the limits on
plasma temperature, density, and confinement time. Here
we explore the range of torque-balanced steady states of
electron plasmas that can be achieved using cyclotron
cooling to balance plasma heating and a rotating electric
field [i.e., the ‘‘rotating-wall’’ (RW) technique], for radial
compression and the mitigation of outward transport.
There is now a large body of knowledge regarding
single-component plasmas in Penning-Malmberg traps
although many questions remain [2]. The rotating-wall
technique was developed to confine and compress electron
[3,4] and ion plasmas [5,6] and was later used to compress
positron plasmas [7], including those used for antihydrogen production [8]. It provides an effective tool to mitigate
outward transport and to compress plasmas radially, across
the magnetic field. With one exception [7], coupling to the
plasma has been achieved by either tuning the RW frequency to a Trivelpiece-Gould (TG) mode [4] or phase
locking to a strongly coupled ion crystal [6]. This Letter
describes a novel, strong-drive RW regime in which efficient compression can be achieved over a broad range of
frequencies without tuning to plasma modes.
Plasma compression is expected to be inhibited by outward transport, typically ascribed to magnetic or electrostatic asymmetries [4,9–11]. The transport rate,
0 1=ndn=dt, is typically found to scale as 0 /
n2 L2p =B2 , where n and Lp are the plasma density and
length, and B is the magnetic field. In the experiments
reported here, plasmas can be compressed to where 0
becomes insensitive to plasma density (i.e., 0 / n0 ).
This regime of reduced transport results in high plasma
densities (e.g., n 1010 cm3 ) with moderately low
plasma temperatures (i.e., in energy units, T & 0:2 eV).
This Letter focuses on describing the range of compressed
0031-9007=05=94(3)=035001(4)$23.00
steady-state plasmas that can be created in this regime. The
implications of these results for the practical limits on
positron storage are also discussed.
Experiments are performed in a cylindrical PenningMalmberg trap, shown schematically in Fig. 1. Electron
plasmas are injected using a standard electron gun, with
initial plasma radii Rp 1–2 mm. Plasmas are confined
radially by the magnetic field, with axial confinement
provided by voltages (typically 100 V) applied to the
end electrodes. The wall diameter is 2.54 cm, and the
plasma length, Lp , can be varied in the range 5 Lp 25 cm, using different electrode configurations.
Rotating electric fields are applied to the plasma using a
special-purpose, four-phase rf generator attached to a foursegment electrode. Each segment extends 90 azimuthally
and 2.54 cm axially; combined, they produce a radial
electric field with azimuthal mode number m 1 rotating
in the same direction as the plasma.
The data presented here correspond to the RW electrode
located near one end of the plasma column, as shown in
Fig. 1, but good confinement and compression have also
been achieved with the RW at other locations.
Theoretically, an m 1 dipole drive applied over the
entire plasma can drive no torque [12]. Although not
studied in detail, we find that good compression can be
achieved as long as the axial extent of the RW electrode is
less than half the plasma length.
The trap is operated in ‘‘inject-manipulate-dump’’
cycles that exhibit very good shot-to-shot reproducibility;
for example, N=N 1%, where N is the particle number.
FIG. 1. Schematic diagram of the experiment (not to scale; not
all electrodes are shown).
035001-1
 2005 The American Physical Society
PRL 94, 035001 (2005)
PHYSICAL REVIEW LETTERS
Dumped plasmas are accelerated to about 5 kV before
striking a phosphor screen, with the resulting images recorded by a charge-coupled device (CCD) camera (effective pixel size 12:5 m in the plasma). This z-integrated
profile and the trap geometry are used as the inputs to a
Poisson solver to calculate the plasma length and density.
For fixed N, the space charge potential, , increases as the
plasma is compressed, and so the plasma length changes
slightly as a function of density. Typical values of range
from 10 to 80 V.
The parallel plasma temperature, Tk , is measured by
slowly lowering the confinement voltage and measuring
the escaping charge [13]. For the plasmas studied here, the
perpendicular-to-parallel equilibration rate is rapid (?k >
1000 s1 ), so Tk T? T [14]. Heat transport is also
rapid with a thermal relaxation time T < 10 ms [15], and
so we assume the temperature is independent of the radius.
The plasma cools by cyclotron radiation in the 5 T magnetic field at a rate c 1=TdT=dt 6 s1 [14,16],
which is fast compared to the compression and expansion
time scales. Thus, in most cases, the plasmas remain
relatively cool (i.e., T & 0:2 eV; T=e 1), even in the
presence of strong RW fields.
After the plasma is injected into the trap, the RW field is
turned on with amplitude VRW and frequency fRW . The
evolution of the plasma is studied by repeating the experiment for different hold times. Examples of plasma compression for Lp 14 cm are shown in Fig. 2(a), with an
applied rotating-wall voltage VRW 1 V at four different
RW frequencies. Here, the RW is on for 20 s, and then the
expansion is studied after the RW is turned off. There are
two data points at each time for each frequency. They
typically differ by <1%, demonstrating the robust compression and reproducibility of the experiments.
As shown in Fig. 2(a), the plasma reaches a steady state
after 5 s. The limiting, steady-state density increases as
the RW drive frequency is increased. Experiments on
longer time scales show that this steady-state density can
FIG. 2. (a) Central density, n0 , vs time during rotating-wall
compression for four different drive frequencies. All cases have
N 5:6 108 , VRW 1:0 V, and Lp 14 cm. (b) Rapid response to an abrupt change in drive frequency: from 6 to 8 MHz
(4), and to 4 MHz (5).
week ending
28 JANUARY 2005
be maintained (n=n < 5%) for more than 24 h. When the
RW is turned off, the density decays exponentially at a
slow rate (< 0:1 s1 ), even at the highest densities studied.
Figure 3 shows three representative radial profiles during plasma compression for fRW 6 MHz. Other than
during rapid compression, the plasma profiles are close
to that of a rigid rotor (i.e., constant density), except
slightly broadened at larger radii. This is likely due to
large plasma viscosity at these densities and temperatures
(viscous relaxation time v < 5 ms) [17,18].
In Fig. 4, the steady-state density is plotted for a broad
range of RW frequencies for three plasma lengths. Both the
density scale and a scale in units of the central E B
rotation frequency, fE , are shown, where fE n0 ec=B 2:9n=1010 cm3 MHz. The thick solid line represents
fRW fE , namely, the density corresponding to rigid
rotation at the RW frequency. The occasional ‘‘steps’’ in
the data are discussed below. Above a minimum RW
amplitude (dependent on frequency, but typically VRW >
0:5 V), strong compression is achieved over a wide range
of frequencies, with no need to tune to a plasma mode. We
refer to this as the strong-drive regime.
At smaller VRW , compression is observed only at distinct
frequencies (e.g., the solid points in Fig. 4), consistent with
previous experiments [4]. However, those experiments
were done at B 4 T with 7 times larger transport rate.
It is likely that the reduced cyclotron cooling and larger
outward transport inhibited observation of the strong-drive
regime reported here.
The data in Fig. 4 illustrate a key result, namely, that, in
the strong-drive regime, the effect of the RW is to increase
plasma density until the plasma rotation frequency is, to
within experimental accuracy, that of the RW. The fact that
the compression ceases at this value of fRW is consistent
with thermodynamic arguments [12]. Further, as shown in
Fig. 2(b), if fRW is abruptly changed after compression to a
steady state, the plasma quickly expands or compresses to
the new steady-state density corresponding to this value of
fRW .
FIG. 3. Three representative profiles during plasma compression for the conditions of Fig. 2, with fRW 6 MHz, at the
times marked by in Fig. 2.
035001-2
PRL 94, 035001 (2005)
PHYSICAL REVIEW LETTERS
FIG. 4. Steady-state density vs applied RW frequency for three
plasma lengths with N 4:5 108 electrons: (4) VRW 1:0 V,
demonstrating strong-drive compression at all frequencies; and
(䊉) VRW 0:1 V for 30 s, showing significant compression only
at discrete frequencies. The solid lines correspond to fE fRW ,
and the dashed lines show the expected density if the RW
frequency matched an m 1 TG mode.
These torque-balanced equilibria are, within limits, not
strongly dependent on the RW amplitude. For example, a
factor of 2 decrease in amplitude (1 to 0:5 V) results in less
than 10% decrease in the steady-state density.
A density limit is observed at n0 3 1010 cm3
(fRW 10 MHz). At higher values of fRW , it becomes
difficult to compress the plasma, even though n0 2:4 104 nB , where nB is the Brillouin density limit. There is
evidence that this is not an intrinsic limit but the result of
transmission-line effects in the rotating-wall circuit, leading to unbalanced drive of the segmented electrodes and
excess plasma heating.
The fact that good compression can be achieved over a
wide range of frequencies and that the plasma evolves
smoothly to a new density limit following an abrupt change
in fRW indicates that RW coupling to the plasma is independent of the low-order TG modes. To within the uncertainty of the data (5%), and ignoring the ‘‘steps,’’ the
strong-drive data in Fig. 4 correspond to fE fRW . If the
coupling involved TG modes, it would be expected that
m fE fRW fp kz Rp =jm mr , with fp the plasma frequency, kz =Lp , and jm mr is the zero of a Bessel
function (see Ref. [4]). The dashed lines in Fig. 4 indicate
the steady-state density expected in this case for m 1,
and mr 1; 2. While we cannot rule out the possibility that
week ending
28 JANUARY 2005
the RW drives higher order modes, we have no evidence
that this is the case.
Shown in Fig. 5 is the expansion rate, 0 , as a function of
plasma density for two different plasma lengths after the
RW is turned off. As discussed above, 0 is typically found
to have a strong dependence on n and Lp . This dependence
has been written in terms of the ‘‘rigidity,’’ R fb =fE , in
the form 0 / R2 / n2 L2 [9,10], where fb v=2L
p is
p
the particle bounce frequency, with v T=m the electron thermal velocity.
The dashed lines in Fig. 5 correspond to a transport rate
0 1:0 s1 n=1010 cm3 2 L=24 cm2 . This is consistent with the best confinement observed in similar electron
plasma experiments [4]. At lower densities, 0 increases as
the square of the density in accord with this scaling. At
higher densities, 0 approaches a maximum and becomes
independent of density. Empirically, this occurs approximately at the point (vertical arrows) at which the collision
frequency ee 3fb for these cold plasmas (i.e., T 0:05 eV with the RW off). When the plasma is highly
collisional, the time for a resonant interaction and
trapped-particle lifetimes is reduced, and so it is plausible
that this will reduce the transport [4].
When the RW is turned off, the plasma expands slowly
while maintaining an approximate rigid-rotor profile. The
exception is very early times (e.g., t < 1 s) where the
central density often drops more rapidly than elsewhere in
the plasma. This may reflect the fact that these RWcompressed, steady-state plasmas are not true thermal
equilibria, and that in steady state, the RW continues to
couple to the central portion of the plasma, even though
fRW fE . It is likely that a theory of these steady states
will require taking into consideration the radial dependences of the RW torque, the viscous relaxation, and the
asymmetry-induced drag on the plasma.
For the data shown in Fig. 2, we can estimate the
minimum RW heating rate, assuming the RW torque,
T RW , is that required to balance the drag on the plasma
measured immediately after the RW is turned off. The
heating power will then be [19]
FIG. 5. Measured outward expansion rate after plasma compression (RW off) for two different plasma lengths. Arrows mark
the densities where ee 3fb , which appears to correspond with
the transition to density independent transport.
035001-3
PRL 94, 035001 (2005)
PH !RW T RW !RW
PHYSICAL REVIEW LETTERS
dL
!RW L0 ;
dt
(1)
P
with L the total plasma angular momentum (L / r2| ,
where r| are the particle radii), and !RW 2fRW .
Balancing this heating with the cyclotron cooling power,
Pc NTc , gives T T TW 15 meV, where TW
is the wall temperature, 25 meV. This estimate yields a
steady-state temperature of T 40 meV, which is a factor
of 5 cooler than that observed. The higher plasma temperature is likely due to excess RW heating.
Previous experiments worked at a lower magnetic field
and a larger transport rate [4], for which Eq. (1) predicts a
factor of 10 higher temperature. This likely prevented
achievement of the high-collisionality and strong-drive
regimes reported here.
The steps observed in Fig. 4 appear to be due to a static
asymmetry in the lab frame that drives a mode rotating
opposite to the plasma, thereby producing a drag. Similar
effects have been observed and studied previously in
strongly coupled ion plasmas subject to laser torques
[20]. The position and extent of these steps are consistent
with low-order Trivelpiece-Gould modes [21], although we
have been unable to make a definitive mode-number
assignment.
In summary, we have discovered a new class of torquebalanced steady states of pure electron plasmas that can be
accessed using strong RW drive and strong cyclotron cooling. Tuning to a plasma mode is not required. The limiting
density corresponds to that at which the plasma rotation
frequency is equal to the applied RW frequency. Plasmas
can be compressed into a regime in which outward
asymmetry-induced transport no longer increases rapidly
with plasma density. The observed radial profiles and
estimates of the thermal and momentum transport and
heating rates indicate that these states are approximately
uniform-temperature rigid rotors.
One important, but as yet unanswered, question is the
maximum density that can be achieved using this technique. Since the observed transport does not increase rapidly with density in this regime, plasma heating will not be
as much of a limiting factor as it would be in the regime
where 0 n2 [22].
The new regime reported here can be expected to be of
considerable practical importance in tailoring antimatter
(e.g., positron) plasmas for a range of applications, including the development of multicell traps for massive and
long-term positron storage [22]. The plasmas studied
here are nearly optimum for such devices. They can be
confined and compressed without tuning to plasma modes,
which facilitates multicell operation. In addition, the reduced transport in the high-density regime reduces plasma
week ending
28 JANUARY 2005
heating and, consequently, the requirements for plasma
cooling.
This work is supported by NSF Grant No. PHY 0354653. The high-field trap was built with ONR support.
We thank J. P. Sullivan and P. Schmidt for their contributions to the design and construction of this device and E. A.
Jerzewski for his expert technical assistance. We acknowledge numerous helpful conversations with F. Anderegg,
C. F. Driscoll, R. W. Gould, R. G. Greaves, and T. M.
O’Neil.
[1] C. M. Surko and R. Greaves, Phys. Plasmas 11, 2333
(2004).
[2] Non-neutral Plasma Physics V, edited by M. Schauer, T.
Mitchell, and R. Nebel (AIP Press, Melville, NY, 2003).
[3] F. Anderegg, E. M. Hollmann, and C. F. Driscoll, Phys.
Rev. Lett. 81, 4875 (1998).
[4] E. M. Hollmann, F. Anderegg, and C. F. Driscoll, Phys.
Plasmas 7, 2776 (2000).
[5] X.-P. Huang, F. Anderegg, E. M. Hollmann, C. F. Driscoll,
and T. M. O’Neil, Phys. Rev. Lett. 78, 875 (1997).
[6] X.-P. Huang, J. J. Bollinger, T. B. Mitchell, and W. M.
Itano, Phys. Rev. Lett. 80, 73 (1998).
[7] R. G. Greaves and C. M. Surko, Phys. Rev. Lett. 85, 1883
(2000); Phys. Plasmas 8, 1879 (2001).
[8] D. P. van der Werf et al., in Ref. [2], p. 172.
[9] C. F. Driscoll and J. H. Malmberg, Phys. Rev. Lett. 50, 167
(1983); C. F. Driscoll, K. S. Fine, and J. H. Malmberg,
Phys. Fluids 29, 2015 (1986).
[10] J. M. Kriesel and C. F. Driscoll, Phys. Rev. Lett. 85, 2510
(2000).
[11] D. L. Eggleston and B. Carrillo, Phys. Plasmas 10, 1308
(2003).
[12] D. H. Dubin and T. M. O’Neil, Rev. Mod. Phys. 71, 87
(1999).
[13] D. L. Eggleston et al., Phys. Fluids B 4, 3432 (1992).
[14] B. R. Beck, J. Fajans, and J. H. Malmberg, Phys. Rev. Lett.
68, 317 (1992).
[15] D. H. E. Dubin and T. M. O’Neil, Phys. Rev. Lett. 78, 3868
(1997); E. M. Hollmann, F. Anderegg, and C. F. Driscoll,
Phys. Plasmas 7, 1767 (2000).
[16] T. M. O’Neil, Phys. Fluids 23, 725 (1980).
[17] T. M. O’Neil, Phys. Fluids 26, 2128 (1983).
[18] J. M. Kriesel and C. F. Driscoll, Phys. Rev. Lett. 87,
135003 (2001).
[19] R. W. Gould, in Non-Neutral Plasma Physics III, edited by
J. J. Bollinger, R. L. Spencer, and R. C. Davidson
(American Institute of Physics, Melville, NY, 1999),
p. 170.
[20] D. J. Heinzen, J. J. Bollinger, F. L. Moore, W. M. Itano,
and D. J. Wineland, Phys. Rev. Lett. 66, 2080 (1991).
[21] J. R. Danielson et al. (unpublished).
[22] C. M. Surko and R. G. Greaves, Radiat. Phys. Chem. 68,
419 (2003).
035001-4